High-level aftereffects reveal the role of statistical features in visual shape encoding

Area of Science:

• Neuroscience
• Computer Vision
• Perception Psychology

Background:

• Visual shape perception is crucial for daily tasks like object recognition and manipulation.
• The neural mechanisms underlying shape encoding in the visual system are not fully understood.
• Visual aftereffects, perceptual distortions after stimulus exposure, offer insights into neural adaptation.

Purpose of the Study:

• To investigate how the human visual system encodes shape.
• To differentiate between low-level and high-level feature contributions to shape representation.
• To understand the role of adaptation in shape perception.

Main Methods:

• Utilized visual aftereffects paradigm to probe shape representations.
• Employed machine learning to synthesize novel shapes in a multidimensional space derived from natural shapes.
• Designed stimuli to create distinct predictions from low-level and high-level adaptation models.

Main Results:

• Adaptation along high-level shape space trajectories better predicted observed shape aftereffects.
• Findings indicate that high-level statistical features play a critical role in visual shape encoding.
• Results suggest human vision is attuned to the statistical regularities of shapes in the natural environment.

Conclusions:

• High-level statistical features are central to visual shape representation.
• Neural adaptation in response to high-level shape properties significantly influences perception.
• The visual system's sensitivity to natural shape distributions is highlighted.